J Comp Physiol A (1991) 169:557 568

Journal of Neural, and Philology

9 Springer-Verlag 1991

Respiratory behavior in the pond snail Lynmaea stagnalis II. Neural elements of the central pattern generator (CPG) N.I. Syed*, and W. Winlow Department of Medical Physiology, University of Leeds, Leeds, LS2 9NQ, England U.K. Accepted August 6, 1991

Summary. Previously (Syed et al. 1991) we described the ventilatory behavior of the pond snail Lymnaea stagnalis and identified motor neurons that innervate various muscles involved in this behavior. In the present study we describe an interneuronal network that controls ventilatory behavior in Lymnaea. An identified interneuron, termed the input 3 interneuron (Ip.3.I), was found to be involved in the opening movement of the pneumostome (expiration), whereas another identified interneuron known as visceral dorsal 4 (V.D.4) caused its closure (inspiration). These cells have reciprocal inhibitory connections with each other, which accounts for their opposing effects on common follower motor neurons. In isolated brain preparations a third identified interneuron, right pedal dorsal 1 (R.Pe.D. 1) initiated the respiratory cycle by the excitation of Ip.3.I. Whereas Ip.3.I in turn excited R.Pe.D.1, the connections between R.Pe.D.1 and V.D.4 were mutually inhibitory. Both Ip.3.I and V.D.4 were active during spontaneously occurring respiratory behavior as recorded from semi-intact preparations, and selective hyperpolarization of V.D.4 during such spontaneous activity disrupted the respiratory behavior. Regarding peripheral feedback, the mechanical stimulation of the pneumostome during its opening movements not only caused closure but also inhibited Ip.3.I in the middle of its discharge. Ip.3.I and V.D.4 were also found to be multifunctional, inhibiting both locomotor and whole body withdrawal neural networks. We conclude from these results that the rhythmic patterned activity underlying respiratory behavior in Lymnaea is generated centrally, and that the network described here therefore comprises a central pattern generator. Key words: Respiration - Interneurons - Neural circuitry Semi-intact-preparation

Peripheral feedback

(and to whom offprint requests should be sent): Department of Medical Physiology, University of Calgary, H.S.C., 3330 Hospital Drive N.W., Calgary, Alberta, T2N 4N1, Canada * P r e s e n t address

Introduction In a companion paper (Syed et al. 1991) we defined various muscles and motor neurons involved in the respiratory behavior of the fresh water pulmonate snail Lymnaea stagnalis. These motor neurons alone are insufficient to account for the sporadic rhythmical nature of the behavior, implying that a higher order neural network must control motor neuron activity. Most rhythmic behaviors of animals, such as locomotion, feeding and respiration have been shown to be under the control of networks of neurons known as central pattern generators (CPGs) (see Delcomyn 1980; Getting 1986, 1988, 1989; Harris-Warrick and Cohen 1985; Kristan 1980; Lydic 1989; Pearson 1985; Selverston 1980). Interactions between the inhibitory and excitatory elements of such a network are thought to underlie the respiratory rhythm of mammals (Wyman 1977; von Euler 1985). Due to the complexity of these networks, very little is currently known regarding the interneurons that comprise the respiratory CPGs of mammals or other vertebrates (see Feldman and Ellenberger 1988). By contrast, in a number of invertebrate preparations, such as locust (Burrows 1975 a, b, 1982) and the mollusc Aplysia (Byrne and Koester 1978; Koester 1989; Alevizos et a1.1989), it has been possible to identify individual interneurons involved in the respiratory behavior of these animals. In the present paper, we describe specific interneurons which control the opening (expiration) and closing (inspiration) movements of the pneumostome (respiratory orifice) of Lymnaea and demonstrate that appropriate connections exist between these interneurons and the relevant motor neurons. Earlier studies carried out on the central nervous system (CNS) of Lymnaea showed that a multiganglionic neuronal ensemble influences various behaviors (McCrohan and Winlow 1985; also see Winlow and Syed 1991). Originally, 3 wide-acting synaptic inputs were described by Benjamin and Winlow (1981), these being termed Inputs 1-3. The pedal giant dopamine cell, now known as right pedal dorsal 1 (R.Pe.D.1) was found

558

to be the source of Input 1, whereas the source of Input 2 was unknown. Similarly, the source(s) of Input 3 (Ip.3), a sporadically occurring compound postsynaptic potential, was unknown although its effects on various follower cells were well characterized (Benjamin and Winlow 1981). In preparations showing no spontaneous Ip.3 activity, it is possible to initiate this activity by electrical stimulation of R.Pe.D1. Winlow et al. (1981) proposed that R.Pe.D.1 caused the activation of Ip.3 interneurons by postinhibitory rebound excitation (PIR). Recently, one of the putative Ip.3 interneurons (Ip.3.I) has been identified (Syed et al. 1990), though its anatomical position renders difficult simultaneous electrophysiological recordings with other known cells (e.g., R.Pe.D.1). Here we present evidence that Input 1 and 3 (but not Input 2) are directly involved in respiratory behavior, the latter by controlling pneumostome opener muscle motor neurons and the former by initiating the respiratory rhythm. Another identified interneuron, described first by Benjamin (1984) as the visceral white interneuron (V.W.I.) and later designated visceral dorsal 4 (V.D.4) by Janse et al. (1985), shares common follower cells with R.Pe.D.1 and Ip.3 interneurons, but its effects on these follower cells are generally opposite to those of Ip.3 (Syed 1988; Syed and Winlow 1989). We provide evidence here that V.D.4 controls pneumostome closure through its effects on the pneumostome closer muscle motor neurons identified in the companion paper (Syed et al. 1991). Furthermore, we show that Ip.3.I and V.D.4 have reciprocally inhibitory connections and that similar connections exist between R.Pe.D.1 and V.D.4. We propose that interactions between at least these 3 interneurons (R.Pe.D.1, V.D.4 and Ip.3.I) underlie the rhythmic respiratory movements of Lymnaea and, as such, constitute a central pattern generator for respiration. We believe the accessibility and simplicity of this 3 interneuron CPG provides an unparalleled opportunity for examining the intrinsic cellular and network properties that underlie rhythmic behaviors. In addition, mechanisms of behavioral integration can be studied since the respiratory CPG of Lymnaea appears to affect various other behaviors, such as locomotion, whole body withdrawal, and heart activity. Materials and methods Specimens of Lymnaea stagnalis (L.) were obtained from animal suppliers (Blades Biological, Sussex, U.K.), maintained at 10-16 ~ C in aerated pond water and fed on lettuce. All experiments were performed on snails of 1-4 g weight. Preparations were bathed in snail saline buffered to pH 7.9 using HEPES (Benjamin and Winlow 1981). Glass microelectrodes were made from circular cross-section (Winlow and Benjamin 1976) filamented glass (internal diameter, 0.69 mm, from Clark-Electromedical instruments) and were pulled using a Searle Bio-Science vertical electrode-puller. The pulled microelectrodes were filled with the saturated solution o f K2SO 4 which gave an electrode resistance of 10-20 MfL For recordings, glass microelectrodes were connected via Ag/AgC1 wires to preamplifiers and the preparation dish was grounded using an Ag/AgC1 bath electrode. Neurolog NL 102G or WP1 M701 preamplifiers were used, both of which have bridge balance cir-

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea cuitry. Amplified signals were displayed on a Tektronix 4 channel oscilloscope. Recordings were made on chart paper using a Gould 4 channel pen recorder. Intraeellular recordings from either isolated brains or semi-intact preparations were made using techniques as described in the companion paper (Syed et al. 1991).

Results

Morphology and electrophysiology of interneurons involved in the respiratory behavior Input 3 (Ip.3) Interneurons. Input 3 is thought to derive from the combined activities of two or more coupled interneurons which have connections with a host of follower cells (Winlow et al. 1981), including R.Pe.D.1 and V.D.4. One of these putative Input 3 interneurons (Ip.3.I) has recently been identified and its synaptic connections with R.Pe.D.1 and V.D.4 have been verified in vitro (Syed et al. 1990; Syed et al., unpublished). Unfortunately, the soma of this cell is situated on the ventral surface of the right parietal ganglion, the surface opposite that of other respiration-associated interneurons and motoneurons (see Fig. 1 of Syed et al. 1991). In semi-intact preparations this anatomical constraint makes it nearly impossible to obtain simultaneous intracellular recordings from Ip.3.I, R.Pe.D.1 and V.D.4. Nevertheless, indirect evidence of Ip.3 interneuron activity could be obtained by recording from known follower cells (e.g., V.J cells) on the dorsal surface of central ganglia. The position of Ip.3.I presented less difficulty in CENTRAL GANGLIONIC RING OF LYMNAEA STAGNALIS DORSAL VIEW

~/ v

.

o

~

Fig. 1. Diagrammatic representation of the central ganglionic ring of Lymnaea showing the location of central neurons used in the present study. Individually identifiable neurons are numbered (e.g., R.Pe.D.I, R.Pe.D.3, V.D.1 and 4, etc.), while identifiable neuronal clusters are given a letter (e.g., L.Pe.F, R.P.A, V.G, etc.) according to the convention of Slade et al. (1981), Kyriakides et al. (1989). The boundaries of the pedal clusters are shown in Kyriakides et al. (1989). I, 2 left and right cerebral ganglia; 3, 4 left and right pedal ganglia; 5, 6 left and right pleural ganglia; 7, 8, left and right parietal ganglia; 9 visceral ganglion; L./R.Pe.E, F left and right pedal E&F cluster neurons; St statocyst; R.P.A group right parietal A group neuron; R.P.D.2 right parietal dorsal 2; V.D.I visceral dorsal 1; V.D.4 visceral dorsal 4; V.E visceral E group; V.G group visceral G group; V.H,LJ, K visceral H, I, J, K cells

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea

559

V.J Cell

I R.Pe.D 1

I 4,e

Ip.3.1 ~

120mY

lOs Fig. 2. Synaptic connections between R.Pe.D. 1, Ip.3.I and a follower V.J cell. In order to expose the ventrally located Input 3 interneuron (Ip.3.1), the right parietal ganglion was twisted around. Simultaneous intracellular recordings were then made from R.Pe.D.1, Ip.3.I and a V.J cell. Electrical stimulation of R.Pe.D.1 (at closed arrow) inhibited the V.J cell whilst exciting Ip.3.I by a biphasic action (i.e. excitation followed by inhibition). Once activated, Ip.3.I excited both R.Pe.D.1 and the V.J. cell

isolated brain preparations, where it could be exposed by twisting the right parietal ganglion. Once exposed, simultaneous intracellular recordings could be made from Ip.3.I and its follower cells. To study the role of Ip.3 activity in respiration we examined the effects of Ip.3 on the V.G group neurons and the V.H,I,J, and K cells, some of which are pneum o s t o m e opener, closer, and mantle cavity muscle m o t o r neurons (see Syed et al. 1991). The locations o f these respiratory neurons and the others used in the present study are shown in Fig. 1. Input 3 excites P.O.M m o t o r neurons (V.J cells) and M.C.M. m o t o r neurons (R.P.A group) while inhibiting P.C.M. m o t o r neurons (V.K cells) (see Syed et al. 1991). Thus, Ip.3 is involved in the opening m o v e m e n t of the p n e u m o s t o m e and therefore in expiration. In the absence of spontaneously occurring Ip.3 discharge it is possible to initiate its activity by electrical stimulation of R.Pe.D.I (Winlow etal. 1981 ; Syed 1988). Once activated by R.Pe.D.1, Ip.3 subsequently excites R.Pe.D.I. We were able to confirm these earlier findings by making direct intracellular recordings from R.Pe.D.1, Ip.3.I and a V.J cell (Fig. 2). In isolated brain preparations, strong electrical stimulation of R.Pe.D1 excited Ip.3.I in a biphasic manner (i.e. inhibition followed by excitation) and once activated, Ip.3.I excited both R.Pe.D.1 and a V.J cell (Fig. 2).

Interneuron visceral dorsal 4 (V.D.4). As already mentioned, Benjamin (1984) identified a neuron located on the dorsal surface of the visceral ganglion which has a variety of effects on a large number of follower cells of the visceral and right parietal ganglia. A neuron morphologically and electrophysiologically similar to this visceral white interneuron (V.W.I.) was described by Janse et al. (1985) as visceral dorsal 4 (V.D.4). Our exten-

B

L+.

I

P

9 10

Fig. 3A, B. Morphology of Interneuron V.D.4. A Photomontage of V.D.4 morphology. Lucifer yellow dye injection revealed V.D.4 as a true interneuron, with no peripheral projections. It has 2 main axons encircling the lower ganglionic ring and has extensive neuritic branches within these ganglia. B Camera lucida drawing based on A, but additional information collected from 5 other preparations has been added. Many V.D.4 neurites are located in the visceral ganglion (V.G.). The interneuron V.D.4 has two almost identical axons encircling paired, parietal, pleural and pedal ganglia. A fine branch was often seen to emanate from the left pleural ganglion (L.P1.G.) which enters the left cerebral ganglion (L.Ce.G.) (open arrows). Both axons of V.D.4 give rise to extensive dendritic fields in all of these ganglia (large arrows). (L left; R right; A anterior; P posterior; St statocyst). Nerves are numbered according to the convention of Slade et al. (1981). Other ganglia labelled: left parietal ganglion (L.P.G.); left pedal ganglion (L.Pe.G); right cerebral ganglion (R.Ce.G.); right parietal ganglion (R.Pa.G.); right pedal ganglion (R.Pe.G.); right pleural ganglion (R.Pl.G) sive investigation of this cell has revealed that V.W.I and V.D.4 are indeed synonymous. In order to be consistent with previously established maps of Lymnaea neurons (e.g., Winlow and Benjamin 1976) we have adopted the nomenclature proposed by Janse et al. (1985). According to Benjamin (1984) V.D.4 ( = V . W . I . ) is a true interneuron; that is, all of its axonal branches are confined within the central ganglionic ring (i.e., it has no

560

N.I. Syed and W. Winlow: Respiratory C.P.G. of

V,K

A

R.P.D2

J

I

Lymnaea

I

V.J Cell

I

113.3.1

40mY

R,P.A 5s

B R.P,D2 ~

~

V.D.

___I 2~176 5s

Fig. 4. Effects of V.D.4 on respiratory motor neurons. Simultaneous intracellular recordings were made from respiratory motor neurons and V.D.4. Electrical stimulation of V.D.4 (at arrows) had inhibitory effects on a pneumostome opener muscle motor neuron (F..J cell) and a mantle cavity muscle motor neuron (R.P.A group), whereas its effects on a pneumostome closer muscle motor neuron (F.K cell) were excitatory. Note: the effects of Ip.3 on all these follower cells are opposite to those of V.D.4 (Syed et al. 1991)

V.D4

Fig. 5A, B. Opposing effects of Ip.3.I and V.D.4 on a common follower cell. Direct electrical stimulation of Ip.3.I (at arrows) had excitatory effects on R.P.D.2 (A) whereas when stimulated, V.D.4 inhibited R.P.D.2 (B)

L.Pe.F

J2OmV

V.D4

J3OmV L.Pe

20mV L.Pe.E . v ~.-_-~:_~"-.,~,

JlOmV 5s

Fig. 6. Effect of V.D.4 activity on locomotor motor neurons. Injection of constant depolarizing current in V.D.4 (at open arrow), caused this interneuron to fire bursts of spikes at regular intervals. These action potentials caused rhythmic inhibitory discharges in L.Pe.E and F cluster neurons. In the L.Pe.E (bottom trace) and

L.Pe.F (top trace) cluster neurons there is a clear 1 : 1 relationship between V.D.4 spikes and the i.p.s.ps (long arrows), whereas V.D.4's inhibitory effect on another L.Pe.F cluster neuron (third trace) is delayed and may be mediated indirectly

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea

561

B Ip.3.1

A

V.D4 lOmV

OmV V.D4 _ ~

.•140mV 5s D

C

V.D4

~

~

/

~

lOs

R.Pe.D1

lOmV

~

5s

~30mV

40mV V

Fig. 7A-D. Direct interactions between Ip.3.I, V.D.4 and R.Pe.D.1 interneurons. Reciprocal inhibitory connections are shown to occur between Ip.3.I and V.D.4 (A, B). The electrical stimulation of either interneuron (at arrow) inhibited the other. A similar inhibitory connection is shown to occur between R.Pe.D.1 and V.D.4 (C, peripheral projections). Our morphological observations confirm these earlier findings. In addition, we find that V.D.4 has extensive axon collateral branches in all central ganglia with the exception of the right cerebral ganglion (Fig. 3 A, B). In contrast to reports by Benjamin (1984), who found this cell to be always silent, in 50% of our fresh preparations (n=220) V.D.4 was spontaneously active, firing bursts of spikes followed by prolonged periods of quiescence or, less commonly, single spikes. In addition to the postsynaptic effects of V.D.4 on the follower cells described earlier by Benjamin (1984) and Janse et al. (1985), many other follower cells have been discovered in the central ganglia of Lymnaea (Syed 1988; Syed and Winlow 1989). The postsynaptic effects of V.D.4 on P.O.M., M.C.M. and P.C.M. motor neurons were found to be opposite to those of Ip.3 (Fig. 4). Briefly, when stimulated electrically, V.D.4 inhibited a V.J cell and a R.P.A group neuron while exciting a V.K cell (Fig. 4). A pair of electrically coupled neurons called visceral dorsal 1 and right parietal dorsal 2 (V.D.1 and R.P.D.2) were previously reported to be involved in respiratory behavior (van der Wilt et al. 1988). We found that Ip.3.I and V.D.4 have opposite effects on these electrically coupled neurons (Fig. 5). We have also found that almost all pedal neurons are follower cells of V.D.4, the major exceptions being giant cells such as left and right pedal dorsal 5,6,7 (L./ R.Pe.D. 5,6,7) (for more details see Syed 1988). Whether

20mV

.

~

|30mV

los D). Either spontaneous or induced action potentials in R.Pe.D.I caused inhibitory postsynaptic potentials in V.D.4 (C). Spontaneous or induced (at arrow)bursts of action potentials in V.D.4 also inhibited R.Pe.D.I (D)

V.D.4 makes monosynaptic connections with all of its follower cells is unknown. However, in the pedal ganglia some of these connections appeared to be monosynaptic on the basis of constancy of delay and one for one occurrence of spikes and postsynaptic potentials (p.s.ps) at all frequencies applied (e.g. Fig. 6). Injection of constant depolarizing current into V.D.4 caused inhibitory rhythmic discharges in pedal locomotor neurons (L./ R.Pe.F) and in neurons which receive synaptic inputs in common with them (e.g., L./R.Pe.E) (Haydon and Winlow 1986), thus preventing normal expression of the locomotor program (Fig. 6). In addition to its effects on locomotor and respiratory motor neurons, V.D.4 excited all V.E group neurons (Syed 1988), including two electrically coupled heart motor neurons described previously by Benjamin et al. (1988). The connections between V.D.4 and Ip.3.I are reciprocally inhibitory (Fig. 7A, B), which is the basis for their opposing effects on common follower cells. Similarly, a mutual inhibitory connection exists between R.Pe.D.1 and V.D.4 (Fig. 7C, D). The inhibitory effects of V.D.4 on R.Pe.D.1 were stronger in those preparations where V.D.4 was spontaneously active (Fig. 7D). As shown previously (Winlow et al. 1981) and also in the present study (Figs. 2, 4, 5), R.Pe.D.1, Ip.3.I and V.D.4 share common follower cells. On these connections, particularly interesting are those of R.Pe.D.1 and Ip.3.I with V.J and V.K cells (Fig. 8). In the absence of Ip.3 activity, R.Pe.D.1 inhibits a V.J cell, and the

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea

562 V.J Cell

I

13~ Ip.3

R.Pe.D1

laomv V,D4

JlOmV V.K Cell

,~

lOs

Fig. 8. Synaptic plasticity of R.Pe.D. 1 connections with its follower cells during the Ip.3 discharge. A depolarizing square pulse of 4 s duration at 15 s intervals was injected into R.Pe.D.1 (at arrow). Stimulation of R.Pe.D.I not only inhibited the V.J cell and V.D.4, and had biphasic effects on a V.K cell, but also switched on Ip.3 discharge (see bar). Both V.D.4 and the V.K cell were hyperpolarized to prevent spiking. Although current injection was maintained

throughout the Ip.3 discharge, R.Pe.D.1 failed to induce its usual effects on its follower cells during Ip.3. Once the powerful discharge of Ip.3 ceased, the physiological contact between R.Pe.D.1 and its follower cells was re-established. It is possible that Ip.3 causes presynaptic inhibition of connections between R.Pe.D.1 and follower cells

V.D.4, and has biphasic effects on a V.K cell (Fig. 8). I n p u t 3 on the other h a n d excites R.Pe.D.1 and V.J cells while inhibiting a V.K cell and the V.D.4. D u r i n g the course o f Ip.3 activity, the postsynaptic effects o f R.Pe.D.1 on V.D.4 and on V.J a n d V.K cells were not observed (Fig. 8). Once the Ip.3 discharge ended, however, the connections between R.Pe.D.1 and these follower cells were restored (Fig. 8). The m e c h a n i s m o f this presynaptic inhibition is at present unclear. However, the possibility o f m o d u l a t i o n o f postsynaptic sensitivity c a n n o t be ruled out.

R.Pe.D1

t~ ILIIII rrrl~rr, V.J Cell

V

. . . . .

V

V.D4

, , ~ J 30mV lOs

Interactions between R.Pe.D.1, Ip.3 and V.D.4 in spontaneously active preparations M o s t o f the results described thus far were obtained f r o m preparations in which interneurons were electrically stimulated and the effects o f this stimulation on follower cells were recorded. The results we n o w describe were obtained f r o m preparations where interneurons R.Pe.D.1, Ip.3.I and V.D.4 were f o u n d to be spontaneously active. These experiments were performed on animals that were kept in 0 2 deprived c h a m b e r s for 24 h. The central neurons f r o m these freshly isolated brain preparations exhibited m o r e s p o n t a n e o u s respiratory patterned activity than could be o b t a i n e d f r o m well aerated animals. Indirect evidence for the occurrence o f Ip.3 was obtained t h r o u g h a previously hyperpolarized follower cell (a V.J cell), whereas direct intracellular recordings were m a d e f r o m hyperpolarized R.Pe.D.1 and

Fig. 9. Respiratory rhythm recorded from an isolated brain preparation. Spontaneously occurring Ip.3 discharges caused the excitation of previously hyperpolarized R.Pe.D.1 and a V.J cell (arrows) while inhibiting V.D.4. Soon after the inhibitory effects of Ip.3, V.D.4 recovered and fired a burst of action potentials and the cycle repeated several times. In this preparation, in order to inhibit their spontaneous activity R.Pe.D.1 and V.J cells were hyperpolarized, therefore the inhibitory effects of V.D.4 on both of these neurons are not obvious V.D.4. W h e n s p o n t a n e o u s l y active, Ip.3 and V.D.4 fired alternating bursts o f action potentials, whereas R.Pe.D. 1 was excited by Ip.3 and inhibited by V.D.4 (Fig. 9).

Electrophysiology of central neurons as recorded from semi-intact preparations Input 3 and pneumostome opening. Muscle tension recordings were m a d e f r o m the opening o f the p n e u m o -

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea

A

563

B

Opening

Pneumostome

-

200mg

i2oOmg

V.D4

Ip.3

V.J C e l l

..... . , , , J 3 o m v

.~120mY lOs

lOs

Closure

C

1200rag

Pneumostome

V.D4 l.p3

Ip.3

m

m

I I

I I

I I

t i

V.J C e l l .._i 20mV

105

Fig. 10A-C. Involvement of Ip.3 and V.D.4 in the pneumostome opening and closure movements. The tension in the pneumostome musculature was recorded during either opening or closure movements, using tension transducers. A In this instance the tension was recorded from the opening of the pneumostome itself. During spontaneously active Ip.3 discharges (bar) a previously hyperpolarized V.J cell was excited by lp.3, which then caused opening of the pneumostome. B The pneumostome was held open by inserting a cotton ball into the opening and then V.D.4 was stimulated electrically. This stimulation (at arrows) induced the closure of the stome during the course of Ip.3 discharge. Again, indirect evidence for Ip.3 activity was obtained from a V.J cell. During these Ip.3 discharges, the pneumostome opened slowly and, after complete distention, remained opened for several seconds (Fig. 10A). Prior to Ip.3 discharge and subsequent pneumostome opening, soap bubbles were placed at the pneumostome. The movements of these bubbles away from the pneumostome indicated that air was being expelled from the lung cavity.

V.D.4 andpneumostome closure. In the absence of Ip.3.I, the opening movements of the pneumostome were induced by inserting a small cotton ball into the orifice. The pneumostome closure movements were then recorded using a tension transducer. The intracellular stimulation o f V.D.4 under these conditions induced closure o f the pneumostome (Fig. 10B). However, this effect was apparently mediated via m o t o r neurons since V.D.4 does not have peripheral projection. During spontaneously occurring respiratory behavior, similar effects

pneumostome as recorded by a tension transducer. C Muscle tension was recorded from the pneumostome closer muscle and intracellular recordings were made from V.D.4 and a V.J ceil. Ip.3 discharge not only excited the previously quiescent V.D.4 following its inhibition, but also the V.J cell (see bars). Soon after Ip.3 activity V.D.4 fired a burst of action potentials, causing the closure of the pneumostome. If V.D.4 was prevented from spiking by imposing a slight hyperpolarization (at small arrow), it did not fire following Ip.3 discharge and the pneumostome remained opened

o f V.D.4 on pneumostome closure movements were observed. Input 3 discharges, as recorded from a V.J cell, excited this follower cell and inhibited V.D.4. Soon after the occurrence o f the Ip.3 discharge, V.D.4 recovered from the inhibition and fired a burst o f action potentials, thus causing the closure of the pneumostome (Fig. 10 C). If V.D.4 was prevented from spiking (through hyperpolarization), it did not fire following Ip.3 discharge and the pneumostome remained opened (not shown here).

Effects of mechanical stimulation of the pneumostome on Ip.3 and its follower cells Mechanical stimulation of the pneumostome area during its opening movement not only caused closure, but inhibited Ip.3 in the middle o f its discharge (Fig. 1 t A). Similarly, the imposed closure o f the pneumostome (by a fine nickel wire attached to the tension transducer) also inhibited Ip.3 in the middle o f its discharge

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea

564 A

Stimulation

Pneumostome

]200mg V.J Cell

Ip.3

Ip.3

Ip.3

C

B

Pneumostome Closed

V.J Cell

1200mg

Pneumostome

Closure

~2

R.Pe.G

Ip.3 V.J Cell 30mV 15s

Fig. 11 A-C. Effects of mechanical stimulation of the pneumostome on Ip.3 and its follower cells. The pneumostome area was mechanically stimulated with a fine brush and the effects of this stimulation on Ip.3 and a V.J cell were recorded. A Tension transducer was attached to the orifice of the pneumostome. Spontaneously active Ip.3 discharge was recorded from a V.J cell (bar) which caused the opening movement of the pneumostome (not shown here). Soon after another Ip.3 discharge the pneumostome was stimulated mechanically (arrow). This mechanical stimulation of the pneumostome not only caused the closure but inhibited Ip.3 in the middle of its discharge. In the absence of mechanical stimulation an uninterrupted Ip.3 discharge was recorded. B The orifice of the pneu-

(Fig. 11 B). Further mechanical stimulation of the pneum o s t o m e area with a fine brush not only inhibited Ip.3 (as recorded f r o m a V.J cell), but also excited a withdrawal m o t o r neuron (R.Pe.G) ( H a y d o n and Winlow 1986), which then induced the whole b o d y withdrawal behavior (contraction of the head-foot complex into the shell) (Fig. 11 C).

Discussion

The experimental results presented here and in the preceding paper (Syed et al. 1991) together f o r m the basis for what we believe is the neural circuitry underlying respiration in the pulmonate snail Lymnaea stagnalis. We described in the c o m p a n i o n paper our observation of respiratory behavior in freely moving animals and identified those m o t o r neurons which are involved in the opening (expiration) and closing (inspiration) movements of the penumostome. In the present paper we have

mostome was sutured closed using fine nickel wire attached to the tension transducer. The spontaneously occurring Ip.3 activity was inhibited in the middle of its discharge due to this imposed closure of the pneumostome. These effects were similar to those of mechanical stimulation of the pneumostome. C Increasing the intensity of mechanical stimulation by use of a blunt glass rod (small black arrow), rather than a fine brush, not only caused the closure of the pneumostome via inhibition of the V.J cell, but also excited a R.Pe.G cluster neuron. The R.Pe.G cluster neuron is a whole body withdrawal motoneuron (Haydon and Winlow 1986). Excitation of this neuron induced whole body withdrawal (open

arrow) identified the relevant interneurons that control the motor output and have examined some of the network properties of the neural ensemble, an ensemble that we believe constitutes a central pattern generator for respiratory behavior.

Interneurons involved in the respiratory behavior The giant dopamine cell. The dopamine neuron (R.Pe.D.1) (Cottrell et al. 1979) has been shown to have well defined effects on its follower cells (Benjamin and Winlow 1981 ; Winlow et al. 1981). The m o r p h o l o g y o f R.Pe.D.1 was revealed by H a y d o n and Winlow (1981) and regeneration studies were carried out by Allison and Benjamin (1985). The work of Benjamin and Winlow (1981) and H a y d o n and Winlow (1981) showed that spikes in R.Pe.D.1 caused excitatory, inhibitory or biphasic postsynaptic potentials (EPSPS, IPSPS, BPSPS) in follower cells present in the visceral and parietal ganglia.

N.I. Syedand W. Winlow: RespiratoryC.P.G. of Lymnaea In semi-intact preparations, after staining with Lucifer yellow dye, we found an axon from R.Pe.D.1 which ramifies through the blood vessels supplying the anterior part of the lung roof. These vessels discharge their blood contents directly into the heart via the vena reno-pulmonalis (Bekius 1972). Electrical stimulation of R.Pe.D.1 in isolated (Fig. 2) and semi-intact preparations (not shown here) initiated Ip.3 activity. This suggests that R.Pe.D.1 is involved in the initiation of the respiratory cycle. How this is brought about in the animal is unclear. However, we hypothesize that R.Pe.D.1 is sensitive to changes in the concentration of oxygen in the blood. A decrease in the blood oxygen would excite R.Pe.D. 1, which in turn may switch on Ip.3 activity by a biphasic response (i.e. excitation followed by inhibition) thus initiating respiratory drive. Further experiments are required to test this hypothesis. As shown earlier, R.Pe.D.1, Ip.3.I, and V.D.4 have a large number of common follower cells on which they have differing effects. Since Ip.3.I and V.D.4 have reciprocally inhibitory connections, their opposing effects on common follower cells are easily comprehended. The connections between R.Pe.D.1 and Ip.3.I, on the other hand, are more complex. R.Pe.D.1 excites Ip.3.I. Once activated, Ip.3.I then excites R.Pe.D.1. In the situation where R.Pe.D.1 and Ip.3.I are activated at the same time (as occurs normally), it is difficult to predict which neuron has priority in commanding the follower system since these two interneurons have opposing effects on some common follower cells. This is particularly problematical if one assumes that all the connections between these interneurons and follower cells are monosynaptic. To define the nature of these connections better, a series of depolarizing square pulses was injected into R.Pe.D. 1. These current pulses excited R.Pe.D.1, which in turn not only had inhibitory, excitatory or biphasic effects on its follower cells, but also initiated the Ip.3 discharge (Fig. 8). It was clear that for the duration of Ip.3 activation R.Pe.D.I failed to induce its effects on follower cells. When the Ip.3 discharge ceased, the physiological interaction between R.Pe.D.1 and its follower cells was re-established (Fig. 8). Thus, Ip.3.I decouples R.Pe.D.1 from its follower cells possibly by presynaptic inhibition or by some other unknown mechanisms. These experiments suggest that during spontaneously active respiratory cycles Ip.3.I, rather than R.Pe.D.1, will dominate the circuitry and perhaps R.Pe.D.I would then act to modulate the motor output. Whichever the case may be, the role of R.Pe.D. 1 in initiating the respiratory cycle by the activation of Ip.3.I is important. This is not surprising since stimulation of biogenic amine-containing neurons has been shown to initiate a number of rhythmic behaviors among both vertebrates and invertebrates (see Harris-Warrick 1988).

Input 3 and pneumostome opening To fully understand the neural basis of rhythmic behaviors, it is first important to understand the endogenous properties of the neurons involved and their network

565 interactions. In cases where the origin of the synaptic inputs is known (e.g., R.Pe.D.1 or V.D.4) it is relatively easy to understand the nature of the connections. With the exception of the neural network underlying feeding behavior (Benjamin and Rose 1985) most of the other wide-acting synaptic inputs and their interactions in the Lymnaea central nervous system, are not well defined. This is because the source(s) of these inputs is/are not known, despite the fact that they are easily observed on various identified cells. Originally these inputs (e.g., Ip.3) were found on numerous cells of unknown function in the parietal and visceral ganglia (Winlow and Benjamin 1976; Benjamin and Winlow 1981). Recently we found Ip.3 discharge to occur on cells of the pedal and cerebral ganglia involved in locomotor and withdrawal behaviors respectively (see Syed 1988). Once we had characterized the motoneuronal function of some of the V.H,I,J,K cells and R.P.A group neurons (Syed et al. 1991), which were originally described as receiving characteristic Ip.3 discharges, we were able to demonstrate the functional significance of Ip.3. Behavioral experiments on the snails suggested that in order to achieve opening movements of the pneumostome the head-foot complex of the animal must be extended out of the shell (i.e., inhibition of whole body withdrawal reflexes) and locomotion must be terminated. When Ip.3 occurred spontaneously it inhibited motor neurons involved in locomotion and withdrawal (Syed 1988), suggesting that Ip.3.I is multifunctional.

V.D.4 and pneumostome closure Janse et al. (1985) and van der Wilt et al. (1987, 1988), showed involvement of V.D.4 in respiratory behavior, correlating its activity to the opening movement of the pneumostome. Our detailed morphological and electrophysiological examinations of V.D.4 (Syed et al., unpublished) have revealed that V.W.I. (Benjamin 1984) and V.D.4 (Janse et al. 1985) are the same cell. Since V.D.4 was always silent in the experiments described by Benjamin (1984), firing bursts of spikes only after electrical stimulation, its interaction with other neurons were speculative. In a majority of the experiments we have described V.D.4 was spontaneously active, but this activity was interrupted by periods of prolonged inhibition. Thus in our preparations it was relatively easy to observe the effects of V.D.4 on follower cells and its interactions with Ip.3 and R.Pe.D.1. The most likely explanation for this difference in spontaneous activity is that in our experiments little or no protease was used to soften the inner sheath (perineurium), whereas in preparations described by Benjamin (1984) protease treatment was carried out for as long as several minutes. In our experience, peptidergic cells such as V.D.4 are more susceptible to proteolytic damage. In most preparations where V.D.4 and Ip.3 were spontaneously active (Fig. 9), reciprocally inhibitory interactions were observed. A burst of spikes in V.D.4 usually followed Ip.3 activity, and vice versa. These connections were confirmed by making direct intracellular recordings from these interneurons. The inhibitory inter-

566 action between Ip.3.I and V.D.4 suggested that these interneurons might have antagonistic motor functions for the same behavior. When V.D.4 was recorded in isolated brain preparations it inhibited P.O.M. and M.C.M. motor neurons, while exciting P.C.M. motor neurons. Similar recordings from semi-intact preparations showed that selective removal of V.D.4 from the circuit by hyperpolarization disrupted respiratory behavior (Fig. 11 C). Like Ip.3 (Syed 1988), V.D.4 was also found to inhibit locomotor and whole body withdrawal motor neurons (Fig. 6) (also see Syed and Winlow 1991). However, the effects of V.D.4 on putative ciliomotor neurons (left and right pedal A-cluster; L./R.Pe.A) were excitatory (Syed et al. 1988; Syed and Winlow 1989), suggesting that following pneumostome closure, V.D.4 initiates the activities of pedal cilia by exciting ciliomotor neurons. The effects of V.D.4 on putative heart motor neurons (Benjamin et al. 1988) were excitatory, perhaps in order to increase cardiac activity so that inspired air can be effectively circulated into the hemolymph (Syed 1988). These effects of V.D.4 on some follower cells which are not involved in respiratory behavior imply that, like Ip.3.I, V.D.4 it is a multifunctional interneuron. The effects of changes in external pOz on the central neurons of Lymnaea stagnalis were described earlier by Janse et al. (1985). These authors and van der Wilt et al. (1987) presented evidence that V.D.4 and visceral E cells were involved in pneumostome opening movements. In addition, a giant neuron present on the ventral surface of the right parietal ganglion (R.P.V.3) was implicated in the closure of the pneumostome (van der Wilt et al. 1987). Considering the fact that V.D.4 is a true interneuron, its effects on pneumostome muscles must be mediated via follower cells. These authors, however, did not address this question since they were unable to identify the appropriate motor neurons. They also failed to observe any interaction between neurons involved in either the opening or the closure of the pneumostome. Finally, no evidence for the direct involvement of these neurons was obtained from semi-intact preparations during spontaneously active respiratory behavior. The results we described here are thus in direct conflict with those of Janse et al. (1985) and van der Wilt et al. (1987, 1988). In addition to the effects of central neurons on pneumostome movements described above, these authors showed that mechanical stimulation of the mantle edge, pneumostome area, and wall of the lung cavity had excitatory effects on right parietal B group (R.P.B group), visceral F group (V.F group) and visceral G group (V.G group) neurons (Janse et al. 1985). We have demonstrated here (Fig. 11), that mechanical stimulation of the pneumostome area not only caused pneumostome closure but also induced whole body withdrawal. Furthermore, all the cell types described by Janse et al. (1985) and van der Wilt et al. (1987, 1988) as receiving excitation during the mechanical stimulation of the pneumostome, lung cavity, and mantle area have been shown here to receive excitation from V.D.4. These findings strengthen our case for the involvement of V.D.4 in closure, rather than opening, of the pneumostome. The reasons for these discrepancies are unclear.

N.I. Syed and W. Winlow: RespiratoryC.P.G. of Lymnaea

A centralpattern generator underlies respiratory behavior in L ymnaea It seems safe to infer that most rhythmic behaviors such as locomotion, feeding and respiration are the outcome of an interaction between CPGs and peripheral feedback (Delcomyn 1980; Kristan 1980; Selverston 1980; Pearson 1985; also see Jacklet 1989, and Getting 1989). The role of peripheral feedback is important for the initiation and termination of respiratory behavior in Lymnaea (see Syed et al. 1991). Nevertheless, the presence of respiratory motor neuronal activity in isolated brain preparations suggests that the basic rhythm is generated centrally. This CPG is composed of at least 3 interneurons (if Ip.3 is considered as one unit): R.Pe.D.1, Ip.3.I, and V.D.4. These interneurons, along with their follower motor neurons, control the pneumostome and mantle cavity muscles. We do not yet know how this rhythm is initiated in semi-intact preparations, but in isolated brain preparations (where Ip.3.I and V.D.4 were quiescent) electrical stimulation of R.Pe.D.1 was able to switch on the respiratory cycle (Syed et al. 1990). Perhaps the same mechanism of activation of the respiratory cycle occurs in intact animals. The respiratory cycle becomes inactivated when the interburst interval between Ip.3 and V.D.4 becomes shorter. Finally, when the two cells burst simultaneously, patterned activity is terminated (not shown).

A proposed circuit for the respiratory oscillator and its interaction with other neural networks Rhythmic activity can also be attributed to an interaction between inhibitory and excitatory rhythm generators. Such a model has been shown to mimic mammalian respiratory rhythms (Bradley etal. 1975; von Euler 1985) and the pyloric rhythm of lobsters (Selverston 1980). The ' half-centre' oscillatory model (Brown 1911) suggests that two populations of neurons with antagonistic motor functions can generate recurrent, cyclical outputs. Reciprocally inhibitory interactions occurring between two cells are the basis for several types of CPGs (see Kristan and Weeks 1983). An extrinsic neuron with common tonic excitation incorporated with two reciprocally inhibitory neurons can form a ring of recurrent inhibition, thus producing a stable oscillatory output that is not self-limiting. In such a network, impulse production is not the function of temporal properties of the neurons, but depends wholly upon their connectivity patterns (Kling and Szekely 1968; Friesen and Stent 1978; von Euler 1985). We propose here that the reciprocally inhibitory interactions between Ip.3.I and V.D.4 closely resemble the 'half-centre' model (Brown 1911). Since both of these interneurons were found to be active only during spontaneously occurring behavior their activities were initiated via other source(s). We showed earlier that in quiescent preparations R.Pe.D.1 initiated Ip.3 activity. In Lymnaea there are some indications that oxygen sensitive receptors exist in the lung cavity (Janse 1981). R.Pe.D.1 is also known to have axonal branches in those blood

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea

567

~L&W.M.N~ P.C.M.N > P.M.N < M.C.M.N~.

[

< P.O.M.N~ J

<

H.M.N

_

ATION"~ I

Ip. 3.1

|

\Pneumo.stome J ~

I

v.u q

I

\Pneumostome/

Fig. 12. A summary diagram showing the interactions between the neuronal elements of the CPG and follower cells. Ip.3 may consist of several interconnected neurons. Stimulation of R.Pe.D.1 excites Ip.3. by a mixed inhibitory-excitatory connection which in turn causes opening of the pneumostome, coupled with expiration. Once activated by R.Pe.D.1, Ip.3 excites R.Pe.D.1 while inhibiting V.D.4. lnterneuron V.D.4 then recovers from the inhibitory effects of Ip.3 and fires a burst of action potentials, causing the closure of the pneumostome. V.D.4 and R.Pe.D.1 have reciprocal inhibitory connections. Both Ip.3 and V.D.4 have inhibitory effects on locomotor and whole body withdrawal motor neurons (L.&W.M.N), whereas the effects of these two interneurons are opposite on pneumostome opener muscle motor neurons (P.O.M.N), mantle cavity motor neuron (M.C.M.N), and pneumostome closure muscle motor neurons (P.C.M.N.). In addition, V.D.4 has excitatory effects on penis retractor motor neurons (P.M.N) and excitatory effects on heart motor neurons (H,M.N) (Syed 1988) vessels which supply the lung area (Syed, unpublished observation). We therefore hypothesize that R.Pe.D.1 is activated by a decrease in the oxygen content in the blood, thus switching on the respiratory cycle. Both Ip.3.I and V.D.4 in turn fire alternate bursts of action potentials, either exciting or inhibiting the appropriate m o t o r neurons. In addition to their effects on respiratory m o t o r neurons, both Ip.3.I and V.D.4 also affect locomotor, whole body withdrawal, and heart m o t o r neurons (Syed 1988). A diagram summarizing these effects is shown in Fig. 12. Our results show that functional connections exist between interneurons and m o t o r neurons comprised within the respiratory C P G of Lymnaea. But even in a relatively simple nervous system such as that o f Lymnaea, it is difficult to demonstrate unequivocally that these neurons make monosynaptic connections and are ' sufficient' ' appropriate' a n d ' necessary' for the respiratory behavior to occur. A better test, therefore, is to attempt to reconstruct the circuit in a culture dish. Using in vitro isolation techniques R.Pe.D.1, Ip.3.I, and V.D.4 were isolated from their respective ganglia and placed in primary cell culture. These neurons exhibited exten-

sive neurite outgrowth and established specific connections with each other (Syed et al. 1990) that were similar to those described in the present study. Furthermore, it was possible to initiate the respiratory cycle by the activation o f R.Pe.D.1 (Syed et al. 1990). These in vitro experiments elucidate and confirm the intrinsic and network properties o f the respiratory neural ensemble described here in a manner unapproachable in either semiintact or isolated brain preparations. In conclusion, our results show that a C P G underlies respiratory behavior in Lymnaea. This C P G is integrative and multifunctional in that it affects several different behaviors. The interneuronal circuitry which controls the inspiratory and expiratory phases of respiratory behavior in Lymnaea has features in c o m m o n with the ' h a l f centre' model whereby two groups o f neurons with antagonistic m o t o r functions reciprocally inhibit each other in a cyclical manner (Brown 1911). This model has been proposed to account for many rhythmic behaviors in both vertebrates and invertebrates (see Kristan 1980), but few direct tests o f the model have been attempted. We believe that further studies of the Lymnaea respiratory C P G will enable us to elucidate some of the principals governing the neural basis of rhythm generation.

Acknowledgements. We thank Dr. R.L. Ridgway, Dr. A.G.M. Bulloch and Ms. K. McKenney for their intelligent and helpful comments during preparation of this manuscript. Thanks are also due to Dr. R.L. Ridgway for photographic assistance and to Mr. D. Harrison for technical assistance. We also thank Caroline Collins for typing the manuscript. This work was supported by a SERC grant to WW., N.I.S. was a University of Leeds Scholar and held on O.R.S. award.

References Alevizos A, Weiss KR, Koester J (1989) SCP-containing R20 neurons modulate respiratory pumping in Aplysia. J Neurosci 9:3058-3071 Allison P, Benjamin PR (1985) Anatomical studies of central regeneration of an identified molluscan interneuron. Proc R Soc B 226:135-157 Bekius R (1972) The circulatory system of Lymnaea stagnalis (L.). Neth J Zool 22:1 58 Benjamin PR (1984) Interneuronal network acting on snail neurosecretory neurons (Yellow Cells and Yellow Green Cells of Lymnaea). J Exp Biol 113 : 165-185 Benjamin PR, Rose RM (1985) Interneuronal circuitry underlying cyclical feeding in gastropod molluscs. In: Edward VE, Steven PW, Bousfield D (eds) The Motor System in Neurobiology. Elsevier Biomedical Press, Amsterdam, pp 67-72 Benjamin PR, Winlow W (1981) The distribution of three wideacting synaptic inputs to identified neurons in the isolated brain of Lymnaea stagnalis (L.). Comp Biochem Physiol 70A:293307 Benjamin PR, Buckett KR, Peters M (1988) Neurons containing FMRFamide-like peptides in the model invertebrate system, Lymnaea. Symp Biol Hung 36:247-259 Bradley GW, Euler C von. Marttila I, Ross B (1975) A model of central and reflex inhibition of inspiration in cat. Biol Cybern 19:105-116 Brown TG (1911) The intrinsic factors in the acts of progression in mammal. Proc R Soc B 84:308 319

568 Burrows M (1975 a) Co-ordinating interneurons of the locust which convey two patterns of motor commands: Their connexions with flight motor neurons. J Exp Biol 63:713-733 Burrows M (1975b) Co-ordinating interneurons of the locust which convey two patterns of motor commands: their connexions with ventilatory motoneurons. J Exp Biol 63:735-753 Burrows M (1982) Interneurons co-ordinating the ventilatory movements of the thoracic spiracles in the locust. J Exp Biol 97: 385-400 Byrne JH, Koester J (1978) Respiratory pumping: Neuronal control of centrally commanded behavior in Aplysia. Brain Res 143 : 87-105 Cottrell GA, Abernethy KB, Barrand MA (1979) Large aminecontaining neurons in the central ganglia of Lymnaea stagnalis. Neuroscience 4:685-689 Delcomyn F (1980) Neural basis of rhythmic behavior in animals. Science 210: 492-484 Feldman TL, Ellenberger HH (1988) Central coordination of respiratory and cardiovascular control in mammals. Annu Rev Physiol 50: 593-606 Friesen WO, Stent GS (1978) Neural circuits for generating rhythmic movements. Annu Rev Biophys Bioeng 7 : 37-61 Getting PA (1986) Understanding central pattern generators. Insights gained from the study of invertebrate locomotion. In: Grillner S, Stein PSG, Stuart DG, Forsberg H, Herman R (eds) Neurobiology of vertebrate locomotion. McMillan Press, London, pp 231-244 Getting PA (1988) Comparative analysis of invertebrate central pattern generators. In : Cohen AH, Rossignol S, Grillner S (eds) Neural control of rhythmic movements. Wiley, New York, pp 101-128 Getting PA (1989) Emerging principles governing the operation of neural networks. Annu Rev Neurosci 12:185-204 Harris-Warrick RM (1988) Chemical modulation of central pattern generators. In: Cohen AH, Rossignol S, Grillner S (eds) Neural control of rhythmic movements. Wiley, New York, pp 285-331 Harris-Warrick RM, Cohen AH (1985) Serotonin modulates the central pattern generator for locomotion in the isolated lamprey spinal cord. J Exp Biol 116:2246 Haydon PG, Winlow W (1981) Morphology of the giant dopaminecontaining neurone, R.Pe.D.1, in Lymnaea stagnalis revealed by Lucifer Yellow (CH). J Exp Biol 94:149-157 Haydon PG, Winlow W (1986) Shell movements associated with locomotion of Lymnaea are driven by a central pattern generator. Comp Biochem Physiol 83A: 23-25 Jacklet JW (1989) (ed) Cellular and neuronal oscillators. Marcell Dekker, New York Janse C (1981) The effects of oxygen on gravity orientation in the pulmonate snail Lymnaea stagnalis. J Comp Physiol 142:51-59 Janse C, Wilt CJ van der, Plas J van der, Roest M van der (1985) Central and peripheral neurons involved in oxygen perception in the pulmonate snail Lymnaea stagnalis (Mollusca, Gastropoda). Comp Biochem Physiol 82A:459-469 Kling V, Szekely G (1968) Stimulation of rhythmic nervous activities. I. Function of networks with cyclical inhibition. Kybernetik 5:89-103 Koester J (1989) Chemically and electrically coupled interneurons mediate respiratory pumping in Aplysia. J Neurophysiol 62:1113-1126 Kristan WB Jr (1980) Generation of rhythmic motor patterns. In: Pinsker HM, Willis WD Jr (eds) Information processing in the nervous system. Raven Press, New York, pp 241-261 Kristan WB Jr, Weeks JC (1983) Neurons controlling the initiation, generation and modulation of leech swimming. In: Roberts A,

N.I. Syed and W. Winlow: Respiratory C.P.G. of Lymnaea Roberts BL (eds) Neural origin or rhythmic movements. Soc Exp Biol Symp 37:243-266 Kyriakides M, McCrohan CR, Slade CT, Syed NI, Winlow W (1989) The morphology and electrophysiology of the neurons of the pedal ganglia of Lymnaea stagnalis (L.). Comp Biochem Physiol 93 A: 861-876 Lydic R (1989) Central pattern-generating neurons and the search for general principles. FASEB J 3 : 2457-2468 McCrohan C, Winlow W (1985) Interganglionic coordination and bilateral symmetry in the nervous system of gastropod molluscs. In: Bush BM, Clarac F (eds) Coordination of motor behavior. Soc Exp Biol Seminar Ser 24:33-62 Pearson KG (1985) Are there central pattern generators for walking and flight in insects? In: Barnes WJP, Gladden MH (eds) Feedback and motor control in invertebrates and vertebrates. Croom Helm, London, pp 307-316 Slade CT, Mills J, Winlow W (1981) The neuronal organization of the paired pedal ganglia of Lymnaea stagnalis (L.). Comp Biochem Physiol 69A: 789-803 Selverston AI (1980) Are central pattern generators understandable? Behav Brain Sci 3:535-571 Syed NI (1988) Neural control of locomotion in Lymnaea. Ph.D. Dissertation, University of Leeds, U.K. Syed NI, Winlow W (1989) Morphology and electrophysiology of neurons innervating the ciliated locomotor epithelium in Lymnaea stagnalis (L.). Comp Biochem Physiol 93A:633-644 Syed NI, Harrison D, Winlow W (1988) Locomotion in Lymnaea: Role of serotonergic A-cluster neurons. Symp Biol Hung 36:387-402 Syed NI, Bulloch AGM, Lukowiak K (1990) In vitro reconstruction of the respiratory central pattern generator of the mollusk Lymnaea. Science 250:282-285 Syed NI, Harrison D, Winlow W (1991) Respiratory behavior in the pond snail Lymnaea stagnalis. I. Behavioral analysis and the identification of motor neurons. J Comp Physiol A 169:541-555 van der Wilt CJ, Roest M van der, Janse C (1987) Neuronal substrates of respiratory behavior and related functions in Lymnaea stagnalis. In: Boer HH, Geraerts WPM, Joosse J (eds) Neurobiology, molluscan models. North Holland, Amsterdam, pp 292-296 van der Wilt CJ, Roest M van der, Janse C (1988) The role of two peptidergic giant neurons in modulation of respiratory behavior in the pond snail, Lymnaea stagnalis. Symp Biol Hung 36:377-386 von Euler C (1985) Central pattern generation during breathing. In: Edward VE, Steven PW, David B (eds) The motor system in neurobiology. Elsevier Biomedical Press, Amsterdam Winlow W, Benjamin PR (1976) Neuronal mapping of the brain of the pond snail, Lymnaea stagnalis (L.) In: Salanki J (ed) Neurobiology of invertebrates, Gastropoda brain. Akademiai Kiado, Budapest, pp 41-50 Winlow W, Benjamin PR (1976) Neuronal mapping of the brain of the pond snail,Lymnaea stagnalis (L.) In : Salanki J (ed) Neurobiology of invertebrates, Gastropoda brain. Akademiai Kiado, Budapest, pp 41-59 Winlow W, Haydon PG, Benjamin PR (1981) Multiple postsynaptic actions of the giant dopamine-containing neuron R.Pe.D.1 of Lymnaea stagnalis. J Exp Biol 94:137-148 Winlow W, Syed NI (1991) Modulation of behavior by a multiganglionic neuronal ensemble. In: Winlow W, Vinogradova O, Sakharov DA (eds) Signal molecules and behavior. Manchester University Press, Manchester (in press) Wyman RJ (1977) Neural generation of breathing rhythm. Annu Rev Physiol 39:417-448